Water tunnels in F1

[volarchico]

Back in the day, I saw a large water tunnel at my university used effectively to model the flow around a 747 model. Granted, this was not up at the Reynolds Number-matched transonic cruise speed, but still. Cavitation was not a problem. Because of the density difference, you can actually run at much slower speeds (or smaller models) than a wind tunnel to get the same results. A fluid being incompressible does not mean it can't be used accurately for features such as diffusers. The density (where the term compressible comes from) does not change that much around an F1 car unless you consider the exhaust.

[RideRate]

Not forgetting the compressibility of water relative to air. That would make it hard to model the air expansion under the floor and diffuser.

There's little expansion under, over, or behind an f1 car. We're well below mach 0.3 and thus incompressible is a valid assumption. The compressibility effects are not dominant and would only warrant investigation after the incompressible flow model is perfected. The air is filling the wake via a high flow rate under the car, not expansion.

[Richard]

Thanks for the feedback.

On a parallel topic, I wonder how they model the effects of the EBD? I guess they must replicate the air jet from the exhaust, presumably with cold air, then try to factor in the expansion/contraction due to the heat?

[gixxer_drew]

This is my first post, but I felt I have some experience on this topic. Air is almost always treated as non compressible for subsonic flow testing. Maybe you could run into something if you were getting transonic in the underbody but the rules really keep the F1 diffusers small, I could be underestimating how well their diffusers are working.

Second, water tunnels have been used for decades for vehicle stuff. AAR for example has one in California, it is a tow type where the model is in motion through a stable medium. The big advantage in my mind of water is to get the reynolds number in scale. It also does an amazing job with flow visualization for things like vortex generators. Hope that helps

-Andrew Brilliant

[Callum]

Welcome to the forum gixxer!

[Pierce89]

Toyota wind tunnel (now used by Ferrari I think) has the capability for PIV, which is a wind tunnel technique that gives full flow field visualisation which can be put in direct comparison with pictures from cfd.
See here:
http://www.tecplot.com/Community/CaseSt ... veeng.aspx
I think for some time toyota's has been the only F1 wind tunnel with such capability (which needs a lot of extra machinery compared to a normal wind tunnel); I do not know if now some other team can do PIV (except from ferrari using toyota's knowledge, of course)

Few teams have tried to implement PIV which is why they have pretty much all gone to Toyota.

PIV is costly, and time consuming to setup. F1 teams do not have time. In the past they have paid for external companies to do such work, coming in, installing their equipment in the respective teams tunnel etc. But as I said, F1 teams can not afford the time or cost for such setups.

Only Toyota could, and only Toyota had the people I believe. However im sure situation has changed and teams do have the capabilities...

With regards to the topic, UK users will im sure heard of a company known as QinetiQ? They are from the aerospace sector. There was a rumour circulating that the Mclaren MP4-25 did indeed have some water channel testing done by these guys.

Water channel testing is not uncommon within the automotive industry due to the Reynolds number effects. Obviously wind tunnels can only run a certain sustainable speed and automotive models can only practically be a certain size meaning more often than not due to cost reasons it can be difficult to achieve complete similarity.... Hence the use of water channels.

With the compressibility, do we have an idea of the flow velocity entering the floor? That is where the peak velocity will be located, but I doubt that velocity will exceed mach 0.3 for most applications where the teams are interested in the flow features. For that reason I would assume compressibility is negligible for these cases.

in 2006 Racecar Engineering did an article saying QinetiQ has a tie-up with Williams. maybe they dumped Sir Frank.

[jon-mullen]

I was taught that compressibility effects are negligible below Mach numbers of 0.3 (103 m/s, 371 kmph, or 230 mph). Is this incorrect?

[PNSD]

in 2006 Racecar Engineering did an article saying QinetiQ has a tie-up with Williams. maybe they dumped Sir Frank.

Im not so sure that still exists.

And Jon-mullen; That indeed is correct. I believe in a copy of John Andersons book of fundamentals of aerodynamics there is a table tabulating the percentage error in calculated force coefficient if you consider incompressible flow analysis for mach numbers of 0 to 1.2? But generally mach 0.3 is considered the cut-off point. In aerodynamics it's a very complicated subject.

I was once taught that I should consider transonic aero effects for mach numbers above 0.7 but then later taught that 0.3 is a far more accurate figure, using the example that during take off and landing manoeuvres a 747 will experience local supersonic flow regions over the wing, yet the freestream mach number may only be 0.3.

[n_anirudh]

...
Second, water tunnels have been used for decades for vehicle stuff. AAR for example has one in California, it is a tow type where the model is in motion through a stable medium. The big advantage in my mind of water is to get the reynolds number in scale. It also does an amazing job with flow visualization for things like vortex generators. Hope that helps
-Andrew Brilliant

Lot of water channels are used with PIV. easier to seed the PIV particles and they can be dissolved using chlorine. I believe this is a bit harder in air..

[PNSD]

Tow tanks are very popular for the automotive industry...

[n_anirudh]

The irony is that with 60% scale model and 250kph top speed F1 wind tunnel do not match real car's Reynolds number for speed greater than 150kph

There are methodics, that increase effective reynolds number.

Do you hin at pressure tempertaure changes to tweak Reynolds or other tricks such as boundary layer tripping?

Well viscosity and density are computed based on a specific time of the day. While doing a normal wind tunnel test, Cl and Cd are monitored. The Reynolds number is later adjusted to map the variations.

[gixxer_drew]

Welcome to the forum gixxer!

Thank you Callum!

On this topic a bit more. I like water tow tunnels for some things but not so much for others. I think every testing type has strengths, weaknesses and various things that are inaccurate in all of them. Even at the level of F1, where they are really on top of what is the most accurate (at any cost!), they are constantly rethinking on this issue. Given enough time to improve it, I believe CFD will become the end all to aerodynamic testing. But that day is not here yet. I had a diffuser design that I changed just one month ago because CFD had told me something completely different than the tunnel. It was worth 2/10th of a second at our 3 min test track! That is an example of how different types of tests can give you different results.

So for now I think the best guys take in all the information they can get, try every type of tunnel and work to learn something from any place you can. Then hope you put it all together in your head and hope you get it right. Some teams get it and some teams are off pace.

I would expect the aero guys in F1 are using all sorts of tunnels.

[Richard]

So for now I think the best guys take in all the information they can get, try every type of tunnel and work to learn something from any place you can. Then hope you put it all together in your head and hope you get it right.

Yup, engineering is rarely ever absolute. A key lesson when I started out was that analysis & testing is informative as to how how something might work. The various combinations and scenarios gives plausible envelopes, but they are probabilistic and we never truly know exactly what is happening.

Obviously multiple methods to analyse/test a problem narrows down the size of that envelope, but it is still only an envelope, and no one method has all the answers.